Previous version of init.d script (cdemud.init.d) had logic for loading vhba kernel module on start and unloading it on stop, but it's gone in current version (cdemud.init.d-1.1.0).
module loading should happen automatically by udev. Probably the following warning has something to do with that: Jan 9 22:52:19 tablet udevd[934]: NAME="%k" is superfluous and breaks kernel supplied names, please remove it from /etc/udev/rules.d/70-vhba.rules:3
Oh, ignore comment #1. Tiziano do you remember why that code was dropped?
It was decided that init.d script should not do that, but it's better to add propper code inside cdemud.
But I suppose we can have this code in init script (as it is already different from upstream one) until this code is in upstream, am I right?
We have a /etc/conf.d/modules for loading modules. I don't see why we need some logic in an apps init.d script.
Well, in general, one doesn't need this module to be loaded the all the time the machine is running. Just for the time cdemud is running. For example, alsasound init script uses similar logic.
(In reply to comment #3) > It was decided that init.d script should not do that, but it's better to add > propper code inside cdemud. > I Agree. Quirking init.d scripts puts too much emphasis on distribution. If this functionality is required, it's best they try to fix it for all distributions, and not just gentoo.
ok, the fix is coming with kernel 2.6.35: http://git.kernel.org/linus/31373d09da5b7fe21fe6f781e92bd534a3495f00
(In reply to comment #8) > ok, the fix is coming with kernel 2.6.35: > http://git.kernel.org/linus/31373d09da5b7fe21fe6f781e92bd534a3495f00 Tiziano, could you clarify, how does laptop-mode per-device flushing helps here?
bah, failed at copy/pasting. here's what I really wanted to paste: http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=578454ff7eab61d13a26b568f99a89a2c9edc881
Created attachment 242713 [details, diff] 1.2.1-devname_kernel-2.6.35.patch A preliminary patch for vhba to make this happen...
(In reply to comment #11) > Created attachment 242713 [details, diff] [details, diff] > 1.2.1-devname_kernel-2.6.35.patch > > A preliminary patch for vhba to make this happen... Doesn't seem to work here (~amd64, kernel 3.1.1, udev-171-r2, module-init-tools-3.16-r1). I have tried applying this patch and using 'KERNEL=="vhba_ctl", MODE="0660", OWNER="root", GROUP="cdemu", OPTIONS+="static_node=vhba_ctl"' in 70-vhba.rules. Now, modules.devname does correctly get a "vhba vhba_ctl c10:250" line, and with the vhba module unloaded, I do get /dev/vhba_ctrl created with the correct permissions after restarting udevd. However, the vhba module does not get loaded when trying to write to /dev/vhba_ctrl. Are there special kernel configs that are required for this to work?
cdemu-daemon does not ship an init.d script anymore